The statistical mechanics of the flux-line lattice in extreme type-II superconductors is studied within the framework of the uniformly frustrated anisotropic three-dimensional XY -model. It is assumed that the externally applied magnetic field is low enough to invalidate the lowest Landaulevel approach to the problem. A finite-field counterpart of an Onsager vortex-loop transition in extreme type-II superconductors renders the vortex liquid phase-incoherent when the Abrikosov vortex lattice undergoes a first order melting transition. For the magnetic fields considered in this paper, corresponding to filling fractions f given by 1/f = 12, 14, 16, 20, 25, 32, 48, 64, 72, 84, 96, 112, and 128, the vortex liquid phase is not describable as a liquid of well-defined field-induced vortex lines. This is due to the proliferation of thermally induced closed vortex-loops with diameters of order the magnetic length in the problem, resulting in a "percolation transition" driven by non-field induced vortices also transverse to the direction of the applied magnetic field. This immediately triggers flux-line lattice melting and loss of phase-coherence along the direction of the magnetic field. Due to this mechanism, the field induced flux lines loose their line tension in the liquid phase, and cannot be considered to be directed or well defined. In a non-relativistic 2D boson-analogy picture, this latter feature would correspond to a vanishing mass of the bosons. Scaling functions for the specific heat are calculated in zero and finite magnetic field. From this we conclude that the critical region is of order 10% of Tc for a mass-anisotropy Mz/M = 3, and increases with increasing mass-anisotropy. The entropy jump at the melting transition is calculated in two ways as a function of magnetic field for a mass-ansitropy slightly lower than that in Y BCO, namely with and without a T -dependent prefactor in the Hamiltonian originating at the microscopic level and surfacing in coarse grained theories such as the one considered in this paper. In the first case, it is found to be ∆S = 0.1kB per pancake-vortex, roughly independent of the magnetic field for the filling fractions considered here. In the second case, we find an enhancement of ∆S by a factor which is less than 2, increasing slightly with decreasing magnetic field. This is still lower than experimental values of ∆S ≈ 0.4kB found experimentally for Y BCO using calorimetric methods. We attribute this to the slightly lower mass-anisotropy used in our simulations. 74.25.Dw, 74.25.Ha,74.60.Ec